Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Wed, 3 Apr 2002 23:01:03 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Wed, 3 Apr 2002 23:00:53 -0500 Received: from tone.orchestra.cse.unsw.EDU.AU ([129.94.242.28]:65432 "HELO tone.orchestra.cse.unsw.EDU.AU") by vger.kernel.org with SMTP id ; Wed, 3 Apr 2002 23:00:41 -0500 From: Neil Brown To: Mike Fedyk Date: Thu, 4 Apr 2002 14:03:16 +1000 (EST) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <15531.53380.136055.226784@notabene.cse.unsw.edu.au> Cc: linux-kernel@vger.kernel.org Subject: Re: Raid5 resync slow with one linear array In-Reply-To: message from Mike Fedyk on Monday April 1 X-Mailer: VM 6.72 under Emacs 20.7.2 X-face: [Gw_3E*Gng}4rRrKRYotwlE?.2|**#s9D Hi, > > I just setup a 4 (5 really) drive raid5 array. > > It is syncing up right now and nothing else is running on the system. > > I have three 18GB SCA scsi drives and 2x9GB linear array in a four "drive" > raid5 array. > > Unfortunately, it is syncing up quite slowly. Only about 2MB/sec on a > 40MB/sec array. The system is idle. > > 2.4.19-pre4-ac3 > > Is there something about this config that says "Don't do that!"? I've > heard about RAID10, but not Linear+RAID5... echo 10000 > /proc/sys/dev/raid/speed_limit_min md tries to monitor the activity on the component devices and limits rebuild activity when there appears to be other activity. It measures "other activity" as "blocks added to kstat.dk_drive_?blk", minus "block due to resync activity". When a component device is an md array, nothing gets recorded in kstat, but lots is recorded as resync activity, so this "other activity" appears as a negative number which, due to storage in an unsigned long, appears rather large. NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/